Breiman, L. (1996b). Bias, variance, and arcing classifiers (Technical Report 460). Statistics Department, University of California. Home/Search Document Not in Database Summary Related Articles Check |
This paper is cited in the following contexts:
First 50 documents Next 50
A Mixed Ensemble Approach for the Semi-Supervised Problem - Dimitriadou, Weingessel.. (Correct)
....runs under a variety of Unix platforms (such as Linux) and under Windows9x 2000 NT. It is available freely via CRAN, the Comprehensive R Archive Network, whose master site is at http: www.R project.org. 3. 2 Data Sets Results Three benchmark are data sets are used for our simulations: twonorm [5], banana [23, 14, 24] and Pima Indians Diabetes [22] Please note, that the same experimental setup is used for every comparisons made. Twonorm is a ## dimensional data set with # classes which are Gaussian clusters. The Bayes error rate is of ####.# data points are used for the training set ....
L. Breiman. Bias, variance, and arcing classifiers. Technical report, Tech. Rep. 460, Statistics Department, University of California, Berkeley, CA, USA, 1996.
A Method to Boost Naïve Bayesian Classifiers - Diao, Hu, Lu, Shi (2002) (Correct)
....boosting) of training sample #. Because we expect the classification results would not be influenced by the order of training sets selected for the base classifiers, the uniform weights suggested by Breiman will be used instead of the weights defined for AdaBoost that depend on the values of # [4][5] Suppose we settle the boosting algorithm to # rounds. For 2 class problems, # should be odd number to avoid ties. If there are more than two classes, ties are possible if some classes have the same number of votes. In this situation, the class predicted by the first classifier with the ....
Bremain, L.: Bias, Variance, and Arcing Classifiers. Machine Learning (2000)
Data Compression and Local Metrics For Nearest Neighbour.. - Ricci, Avesani (1997) (2 citations) (Correct)
....when k is optimized with cross validation. LASM introduces several new parameters, namely the weights of the local metrics. This suggests that LASM improves 1 NN accuracy by reducing the bias component of the error but increasing the variance component. In these cases, as others have shown [4], the prediction accuracy can be improved by making voting different versions of the same classifier. 11 worse (better) than V LASM for at least at 0.02 level for a paired t test. DATA LASM V LASM k NN Ba 85.7 Sigma2.1 90.1 Sigma1.9 87.9 Sigma2.3 Wi 73.5 Sigma4.4 76.0 Sigma4.0 ....
....the prototypes is more likely to produce different ET sets. In our approach voting always improves or equate the accuracy of the k NN classifier (not only the 1 NN) On small dimensional data sets (Echocardiogram, Thyroid and Iris) voting does not improves LASM because only a small instability [4] is introduced by the additional parameters brought in by the local metric. And it is known that voting can only reduce the variance component of the error in unstable classifiers. 7 Conclusion and Future Directions In this paper we have studied a weighting scheme for the nearest neighbor ....
L. Breiman. Bias, variance, and arcing classifiers. Technical Report 460, University of California, Berkeley, April 1996.
The Alternating Decision Tree Learning Algorithm - Freund, Mason (1999) (22 citations) (Correct)
....cost of abstaining from predicting examples that are hard to classify. 1 INTRODUCTION The AdaBoost algorithm [7, 16] has recently proved to be an important component in practical learning algorithms. Two of the most successful combinations have been boosting decision trees and boosting stumps [6, 1, 13, 8] . Stumps are the simplest special case of decision trees which consist of a single decision node and two prediction leaves. Boosting decision trees learning algorithms, such as CART [2] and C4.5 [14] yields very good classifiers. The software package C5.01 is one of the best packages that ....
Leo Breiman. Bias, variance, and arcing classifiers. Technical Report 460, Statistics Department, University of California at Berkeley, 1996.
Error-Correcting Output Coding for Text Classification - Berger (1999) (19 citations) (Correct)
....types of classifiers to the same problem [ Quinlan, 1993 ] Why consensus algorithms work so well in practice is still an open question. As a step in that direction, theoretical work has recently established that combining multiple runs of a classification algorithm can reduce its variance [ Breiman, 1996b ] Unlike most voting algorithms, the constituent classifiers in error correcting output coding aren t all solving the same problem; in fact, they are each solving a distinct binary classification problem. Kong and Dietterich, 1995 ] have shown that this property of the ECOC algorithm bestows ....
L. Breiman. Bias, variance, and arcing classifiers. Technical report, Statistics Department, Stanford University TR-460, 1996.
The mlbench Package - December Topics Documented (Correct)
.... and Andreas Weingessel Examples p mlbench.cuboids(7000) plot(p) library(xgobi) xgobi(p x,color=c( black , white , red , green ) p cl] 22 mlbench.friedman2 mlbench.friedman1 Benchmark Problem Friedman 1 Description The regression problem Friedman 1 as described in Friedman (1991) and Breiman (1996). Inputs are 10 independent variables uniformly distributed on the interval [0, 1] only 5 out of these 10 are actually used. Outputs are created according to the formula y = 10 sin(#x1x2) 20(x3 0.5) 2 10x4 5x5 e where e is N(0,sd) Usage mlbench.friedman1(n, sd=1) Arguments n ....
.... 20(x3 0. 5) 2 10x4 5x5 e where e is N(0,sd) Usage mlbench.friedman1(n, sd=1) Arguments n number of patterns to create sd Standard deviation of noise Value Returns a list with components x input values (independent variables) y output values (dependent variable) References Breiman, Leo (1996) Bagging predictors. Machine Learning 24, pages 123 140. Friedman, Jerome H. 1991) Multivariate adaptive regression splines. The Annals of Statistics 19 (1) pages 1 67. mlbench.friedman2 Benchmark Problem Friedman 2 Description The regression problem Friedman 2 as described in Friedman (1991) ....
[Article contains additional citation context not shown here]
Breiman, L. (1996). Bias, variance, and arcing classifiers. Tech. Rep. 460, Statistics Department, University of California, Berkeley, CA, USA.
Progress Report: Ensemble Methods in Connection with Neural .. - Jakob Vogdrup Hansen (Correct)
....deterministic with very simple rules, was extended to a simulator with artificial intelligence using genetic programming and genetic algorithms [12] Interesting behavior was found. I read some articles about ensemble methods and decided to concentrate on that. The most motivating articles were [4], 1] and [9] In 1997 I developed an alternative ensemble method, called the DynCo Ensemble Method. I wrote an article [13] for a special issue of Machine Learning, which is in 4 submission at this time. It is due for publication in March 1998 and can be found in appendix C. 3 Introduction to ....
....R R f(x)p(x)dx where p(x) is the probability density, while the mean for a discrete function hf(x)i x is P f(x)P (x) where P (x) is the probability. 4. 2 Bias Variance The notion of bias and variance in connection with generalization error and learning is discussed many places e.g. 10] and [4]. Let the error of a predictor f at x be defined as the square error 1 2 (t(x) Gamma f(x) 2 where t is the function we want to learn. The generalization error is: E = h 1 2 (t(x) Gamma f(x) 2 i x (2) The predictor is made using a training set D. In order to emphasize this, the ....
Breiman, L. Bias,Variance, and Arcing Classifiers. Tech. Rep. 460, Statistics Department, University of California, Berkeley, CA 94720, Apr. 1996.
Dynamic Coefficients in Neural Network Regression Ensembles - Jakob Vogdrup Hansen (1998) (Correct)
....purpose of generating new and diverse training sets. A group of predictors are then trained on these new training sets. The result is that the variation in the predictors will give a positive effect on generalization by reducing the variance term in the error of the combined predictor 1 . see [3] or section 3 on bias variance) One result for regressors (e.g [1] states that an ensemble never can do worse than the mean of the predictors. A few of the ensemble methods are Bagging [2] and AdaBoost [7] To combine the members of an ensemble, different techniques have been used. For ....
Breiman, L. Bias,Variance, and Arcing Classifiers. Tech. Rep. 460, Statistics Department, University of California, Berkeley, CA 94720, 1996.
Pruning in Ordered Bagging Ensembles - Martinez-Munoz, Suarez (2006) (Correct)
No context found.
Breiman, L. (1996b). Bias, variance, and arcing classifiers (Technical Report 460). Statistics Department, University of California.
Switching Class Labels to Generate Classification Ensembles - Martínez-Muñoz, Suárez (2005) (Correct)
No context found.
L. Breiman, Bias, variance, and arcing classifiers, Tech. Rep. 460, Statistics Department, University of California (1996.
Pruning in Ordered Bagging Ensembles - Suarez (2005) (Correct)
No context found.
Breiman, L.: Bias, variance, and arcing classifiers. Technical Report 460, Statistics Department, University of California (1996)
Aggregation Ordering in Bagging - Martínez-Muñoz, Suárez (2004) (Correct)
No context found.
L. Breiman, Bias, variance, and arcing classifiers, Technical Report 460, Statistics Department, University of California, 1996
Using Boosting to Prune Bagging Ensembles - Suarez (Correct)
No context found.
Breiman, L., 1996b. Bias, variance, and arcing classifiers. Tech. Rep. 460, Statistics Department, University of California.
On The Size of Training Set and - The Benefit From (Correct)
No context found.
Breiman, L.: Bias, variance, and arcing classifiers. Technical Report 460, Statistics Department, University of California, Berkeley, CA (1996)
Ensembles of Multi-Instance Learners - Zhou, Zhang (2003) (2 citations) (Correct)
No context found.
Breiman, L.: Bias, variance, and arcing classifiers. Technical Report 460, Statistics Department, University of California, Berkeley, CA (1996)
Soft Margins for AdaBoost - Rätsch, Onoda, Müller (1998) (Correct)
No context found.
L. Breiman. Bias, variance, and arcing classifiers. Technical Report 460, Statistics Department, University of California, July 1997.
Boosting in the limit: Maximizing the margin of learned ensembles - Adam Grove And (1998) (75 citations) (Correct)
No context found.
L. Breiman. Bias, variance, and arcing classifiers. Technical report, Statistics Department, U. C. Berkeley, 1996.
A Constructive Algorithm for Training Cooperative Neural.. - Islam, Yao, Murase (2003) (2 citations) (Correct)
No context found.
, "Bias, Variance, and Arcing Classifier," Univ. California -Berkeley, Berkeley, CA, Tech. Rep. 460, 1996.
Diversity Creation Methods: A Survey And Categorisation - Brown, Wyatt, Harris, Yao (2004) (3 citations) (Correct)
No context found.
L. Breiman, Bias, variance, and arcing classifiers, Tech. Rep. 460, Statistics Department, University of California, Berkeley (1996).
Diversity in Neural Network Ensembles - Brown (2003) (1 citation) (Correct)
No context found.
Leo Breiman. Bias, variance, and arcing classifiers. Technical Report 460, Statistics Department, University of California, Berkeley, 1996.
Ensembles of Multi-Instance Learners - Zhou, Zhang (2003) (2 citations) (Correct)
No context found.
Breiman, L.: Bias, variance, and arcing classifiers. Technical Report 460, Statistics Department, University of California, Berkeley, CA (1996)
Diversity in Neural Network Ensembles - Gavin Brown To (2003) (1 citation) (Correct)
No context found.
Leo Breiman. Bias, variance, and arcing classifiers. Technical Report 460, Statistics Department, University of California, Berkeley, 1996.
Soft Margins for AdaBoost - Gunnar Ratsch Gmd (1998) (Correct)
No context found.
L. Breiman. Bias, variance, and arcing classifiers. Technical Report 460, Statistics Department, University of California, July 1997.
CGBoost: Conjugate Gradient in Function Space - Li, Abu-Mostafa, Pratap (2003) (Correct)
No context found.
L. Breiman. Bias, variance, and arcing classifiers. Technical Report 460, Department of Statistics, University of California at Berkeley, Apr. 1996.
On The Size of Training Set and The Benefit from Ensemble - Zhou, Wei, Li, Dai (2004) (Correct)
No context found.
Breiman, L.: Bias, variance, and arcing classifiers. Technical Report 460, Statistics Department, University of California, Berkeley, CA (1996)
First 50 documents Next 50
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC